To identify all the types of financial incentives that have been provided to health care professionals and, when possible, to assess the effects of these incentives on the costs, process or outcomes of health care. The financial incentives considered were those that had been proposed, described, or used regardless of their initial objective.

MEDLINE, EMBASE, the Health Planning and Administration database, Pascal, International Pharmaceutical Abstracts, and the Cochrane Library were searched from January 1993 to May 1999 for English and French publications . The extensive list of search terms provided, included the following: 'physician practice patterns', 'physician behaviour', 'physician attitude', 'physician and clinical practice', 'medical care', 'patient care', 'health care', 'health care delivery', 'health care utilization', 'laboratory tests', 'prescriptions', 'decision making', 'physician incentive plans', 'economics', 'information systems', 'health planning', 'organization and administration', 'quality assurance', 'guidelines', 'feedback', 'drug industry', 'leadership', 'education' and 'reminders'. The authors also reported search terms referring to study design and research type (e.g. pilot projects). Additional papers were retrieved from the bibliographies of selected articles.

Study designs of evaluations included in the review

The authors searched for randomised controlled trials (RCTs), controlled clinical trials, clinical trials, prospective studies, retrospective studies, longitudinal studies, evaluation studies, programme evaluations, health services research, intervention studies, pilot projects, and comparative studies. In the review, the authors discussed the studies as RCTs, same-physician studies, same-patient studies, or same-disease studies.

Specific interventions included in the review

Financial incentives for health care professionals. These incentives included: 'fund-holding' (capitated payment for each patient registered), 'fee-for-service', and 'salary'. In addition, these incentives and others were examined within, or compared with, various managed care environments (health insurance, children's Medicaid programme, and Medicaid health management organisation). The authors stated that studies were excluded if there were bilateral agreements between the physicians and the pharmaceutical industry.

Participants included in the review

The authors did not report any inclusion or exclusion criteria relating to the participants. The study participants appeared to be largely comprised of physicians, patients or insurance carriers.

Outcomes assessed in the review

The primary outcomes assessed were the costs, the use of health care services, the process of care and the outcome of care. There were numerous outcomes reported under each of these headings. The authors described studies in Europe and Canada separately from those conducted in the USA, since studies within the USA operated in a managed care environment.

For studies conducted in Europe and Canada, the outcomes included: general practitioner (GP) workload; the prescription costs; the number of drugs per prescription; the number of referrals for elective surgery or for private clinics; the number of prescriptions and the number of prescriptions for generic drugs; the number of GP visits at night; and the number of elective procedures.

For studies conducted in the USA, the outcomes reported under the heading of 'use of health care resources' included: the reduction in patient costs, e.g. due to length of stay in hospital and ambulatory costs; hospital admission rates; the number of physicians' visits; the number of emergency visits; physicians' recommendation of services; and the number of prescriptions. The study outcomes reported under the heading of 'process of care' included: the number of diagnostic procedures; the quality of clinical decision-making; the number of procedures performed; hospital management of patients; compliance with practice guidelines; rate of immunisation; the amount and quality of treatment; mortality; and screening practices.

The study outcomes reported under the heading of 'outcomes of care' included: access to care, quality of care, patient satisfaction, and general health rating scores.

How were decisions on the relevance of primary studies made?

Two reviewers independently reviewed the citations and abstracts, with nomination by any one reviewer leading to retrieval of the full-length article. Any disagreements were resolved by discussion.

The quality of each study was assessed according to the criteria described by the Cochrane Effective Practice and Organisation of Care Group (see Other Publications of Related Interest). The full-length papers retrieved were examined by two reviewers together, and any disagreements were resolved by discussion.

The full-length papers retrieved were examined by two reviewers together, and any disagreements were resolved by discussion. For RCTs only, the data were extracted for the categories of: sample population, i.e. physicians, insurance carriers or participants; sample size; intervention; results; and study identification.

How were the studies combined?

A narrative synthesis was undertaken, where the studies were grouped and discussed according to the following categories: fund-holding and fee for service and salary, for those studies conducted in Europe and Canada; use of health care resources, process of care and outcomes of care, for those studies conducted in the USA.

How were differences between studies investigated?

The authors stated that they could not conduct a meta-analysis due to the small number of randomised trials, and the lack of comparability for both intervention type and study populations (physicians or patients). Studies conducted in the USA were described separately from those conducted in Europe and Canada, since most of te US studies examined financial incentives within a managed care environment. In addition, the Canadian and European studies were categorised by the type of financial incentive, while the US studies were categorised according to the broad outcome variables.

Eighty-nine studies were reported to have been 'kept for final analysis', of which 8 were RCTs; the authors did not report the numbers of the other study designs included. The actual number of studies that addressed the review question appeared to be closer to 36 (including the RCTs). There were 5 RCTs involving 1,446 physicians in total, 1 RCT of 24 insurance carriers, and 2 RCTs involving a total of 8,245 patients. The number of participants was not reported for the other included studies.

The authors identified several types of financial incentives from a review of the literature. They also identified a number of confounding factors and risks relating to the use of financial incentives. The confounding factors included: the age of the doctor; experience and qualification; place and type of medical practice; previous sanctions for over-prescribing; type and severity of disease; and type of insurance. The risks of financial incentives were based on limited access to certain types of care, lack of continuity of care, and conflicts of interest between the physician and the patient. The authors stated that only a few studies met the basic quality criteria, and that the results of the studies were often preliminary. They also stated that few studies used the same methodology to assess the impact of the same incentive, thus limiting the external validity of their conclusions. The general results appeared to be similar when comparing the effects of financial incentives within a managed care environment (i.e. US studies) to those without (i.e. European and Canadian studies): salary and capitation or fund-holding reduced the total volume of prescriptions by 0 to 24%, and hospital days by up to 80%, compared with fee-for-service. An annual cap on the doctors' incomes resulted in referrals to colleagues when the target income was reached.

Brief details were reported on some cost outcomes. One study reported that managed care had been found to reduce the cost per admission by on average 20%. Greater reductions were found in those hospitals that promoted the sharing of resource use with physicians and disseminated guidelines on the processes of care.

Financial incentives can be used to reduce the use of health care resources, improve compliance with practice guidelines or achieve a general health target. It may be effective to use combinations of incentives, depending on the target set for a given health care programme.

The review question was broad as the research was intended to be exploratory. Details of the inclusion or exclusion criteria relating to study design, participants, intervention, and outcome of interest, were therefore lacking. The authors stated that 89 of 130 studies 'met the criteria described', but this only referred to the 'type of incentives, results of the incentive on selected end-points, and quality of methodology'. The authors should have described clearly what incentives were included, what end points were selected, and what level of quality had to be met. In addition, the authors made no distinction between those papers that contributed general information to the review and those that reported on an intervention and outcome. It was stated that 89 papers were included in the review, whereas only 36 appeared to directly address the review question. The authors did, however, report that the abstracts and citations were independently reviewed, with nomination by any one reviewer leading to retrieval of the full-length article.

The literature search was adequate with a number of databases being searched. However, the search was restricted to publications in English or French, and no effort was made to locate unpublished literature or contact experts in the field. A lengthy list of search terms was provided, but this did not include terms relating to specific financial incentives or outcomes of interest. It is likely, therefore, that some studies were missed.

The authors stated that the quality of the papers was assessed, although the results of this assessment were not reported. Moreover, it is not entirely clear whether all the papers included in the review actually met the quality criteria; the authors stated in the materials and methods section that papers were selected and analysed by quality of methodology, yet also stated in the discussion that 'few' met the quality criteria. It is unclear if this 'few' referred to the 36 papers discussed. The authors reported that the full-length papers were reviewed by two researchers together, rather than independently. Insufficient information was presented on the included studies, with most of the studies being only briefly described in the text. There was a table providing more detail on the eight RCTs, but some information was still lacking, e.g. sample size and characteristics of the intervention and control groups.

The studies were summarised appropriately as a narrative synthesis. However, due to the large number of heterogeneous studies, the conclusions do not appear to be fully substantiated, and should be treated with caution. The authors stated in the discussion that the review was only preliminary.

Practice: The authors state that financial incentives can be used to reduce the use of health care resources, improve compliance with practice guidelines or achieve a general health target. The implementation of financial incentives in the health care system should be simple, transparent and direct: there should be a binary relationship between the incentive and the desired behaviour from doctors or patients.

Research: The authors do not state any implications for further research.

French Ministry of Health.

Bero LA, Grilli R, Grimshaw JM, Mowatt G, Oxman AD, Zwarenstein M, editors. Cochrane Effective Practice and Organisation of Care Group. In: The Cochrane Library, Issue 1, 2002. Oxford: Update Software.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.